OpenAPI rate limits

The KakaoCloud OpenAPI server operates a rate limit policy to provide stable and reliable API services to all users. When excessive requests are generated by a specific client or system, requests can be limited to prevent system resource exhaustion and maintain high availability across the service.

This document explains default quotas, limit scopes, responses when quotas are exceeded, and recommended client error handling.

Default quotas

KakaoCloud OpenAPI applies the following default policies together without additional configuration. Limits are counted independently for each identification scope.

• Multi-layer limits: Requests must satisfy all DOMAIN, PROJECT, IP, and USER limits. If any scope exceeds its threshold, the request is blocked.
• Independent counting: For example, an increase in requests from User A does not affect the request count for User B in the same project.

Limit scope	Per second	Per minute	Adjustable
Domain (DOMAIN)	1,000 requests	5,000 requests	Yes
Project (PROJECT)	100 requests	3,000 requests	Yes
IP address (IP)	30 requests	600 requests	Yes
User (USER)	10 requests	300 requests	Yes

The listed quotas are commonly applied to all users for system stability. If you need a quota adjustment, contact us through the Helpdesk.

Response and error handling

When a request is blocked because it exceeds the configured quota, the API server returns the HTTP response code 429, status message Too Many Requests, and an error message.

Clients that receive a 429 response are recommended to apply the following best practices.

• Retry with exponential backoff: Do not send immediate consecutive requests. Double the wait time between retries.
• Add jitter: Add a random delay to the retry wait time to prevent many clients from concentrating requests at the same time, also known as the thundering herd problem.
• Distribute requests: For batch jobs or bulk calls, throttle request rates so they stay within the per-second and per-minute limits.